Background

Currently Latinxs are over 18% of the US population and make up over half of the total population growth in the past decade [1]. This population is composed of foreign- and US-born peoples who can trace their descent to over 30 countries [1]. Despite the large size of the Latinx population in the US, public health research often analyzes this heterogeneous population as a monolithic group and, perhaps as a result, the data acquired by researchers studying the mental health of this group has sometimes yielded inconsistent or puzzling findings [2,3,4,5,6].

Studies which do not disaggregate the US Latinx population by ethnicity suggest that this population has better mental health than non-Latinx whites, despite having disadvantaged socioeconomic status and less access to healthcare [8,9,10,11,12]. Further studies, which disaggregate the Latinx community by nativity, indicate that immigrants report significantly lower rates of mental illnesses than their US-born counterparts [13]. Scholars have theorized that a “healthy immigrant effect” resulting from the ability of healthier immigrants to migrate and the probability that unhealthier immigrants emigrate back might explain these associations [14]. However, studies which disaggregate among Latinx immigrant ethnicities indicate that significant heterogeneity exists in the rates of mental illness and mental health trajectories over time spent in the US [4,5,6,7].

Studies using the 2001–2003 waves of the National Comorbidity Survey Replication (NCS-R) and the National Latino and Asian American Study (NLAAS) found that Latinx subjects in aggregate (immigrants and US born) reported lower age-, gender-, socioeconomic-adjusted lifetime prevalence rates for any depressive disorder (15.4%) and any anxiety disorder (15.7%) compared to non-Latinx whites (22.3% and 25.7%, respectively) [12]. Further disaggregation indicated that among Mexicans, Puerto Ricans, and Cubans, Puerto Ricans (followed by Cubans) reported the highest lifetime prevalence rates of any depressive disorder (37.4%) and any anxiety disorder (21.7%) and reported no significant protective effect of foreign-born nativity. Findings from the National Epidemiological Survey of Alcohol and Related Conditions (NESARC) came to similar conclusions [5].

The studies mentioned thus far measure mental wellbeing by estimating prevalence rates as the presence or absence of psychiatric disorders using structured diagnostic instruments based on either the DSM-IV or the DSM-V. Although the NLAAS took several steps to attain cultural relevance, there is little consensus on the effectiveness of cross-cultural measurements of these disorders [15]. In fact, it is often self-assessment, through an individual’s perception of their own mental health, that prompts the decision-making process to seek mental health care [15,16,17,18,19,20].

Theoretical/Conceptual Framework

This study builds on the work of critical race scholars and decolonial scholars to offer an extension of the racialized ethnicities framework and locate the differences in mental health outcomes among Latinx migrants within a broader context of core-periphery sociopolitical power relations between their country of origin and the United States [21]. Ramon Grosfoguel’s framework proposes a tripartite classification system composed of ‘immigrants,’ ‘colonial immigrants,’ and ‘colonial racialized subjects.’ He argues that these groups experience different processes of ethnic group incorporation and racialization as a result of different histories of colonial relations and contemporary core-periphery economic relations between the state of origin and the host state [21].

I extend this theorizing to propose an understanding of the differential health outcomes among Latinx immigrants, as a result of this gradient of exposure to US colonialityFootnote 1 and the psychosocial harms associated with the consequences of incorporation as colonial racialized subjects. This gradient of exposure ranges from the least exposure among migrants from states without recent histories of overt colonialism by the US, who are categorized as ‘immigrant’ within Grosfoguel’s typology, to the most exposure among migrants from current US colonial administrationsFootnote 2, who are categorized as ‘colonial racialized subject’ within Grosfoguel’s typology. I theorize that migrants from states with neocolonial relationships and migrants from states with recent colonial administrations would hold middling positions in this gradient.

The goal of this study is to examine self-reported feelings of depression, anxiety and psychological distress among immigrants from Mexico, Puerto Rico, Cuba, the Dominican Republic, and an aggregate group of countries in Central and South America. Following the predicted typology, I hypothesize that membership in the Dominican and Cuban ethnic groups (due to the history of US military intervention and occupations in these countries in the early 20th century), as well as membership in the Puerto Rican ethnic group (due to Puerto Rico’s status as a non-sovereign territory under US colonial administration), will report the highest rates of each measure of distress. I hypothesize that Mexican migrants (due to a history of settler colonial land occupation by the United States in early 19th century Mexico) will report intermediate rates of distress, and Central and South Americans (due to neocolonial relationships with the US) will report the lowest rates of distress. This study updates prior research by using the most recent wave of data from the National Health Interview Survey (NHIS), and addresses potential cross-cultural measurement error in diagnostic criteria, by examining subjective mental health rather than diagnosable symptomology.

Methods

Data

The data for this analysis were acquired from IPUMS National Health Interview Survey (NHIS) (whose questionnaire is administered in both English and Spanish) for the years 2010–2018. The sample is N = 10,465 observations after removing missing data and selecting the subpopulation of individuals who self-identify as Latinx and chose a country of birth outside of the US. The distribution of the weighted subpopulation sample reflects that of the greater Latinx population in the US, with 56% identifying as Mexican, 8% as Puerto Rican, 6% as Cuban, 5% as Dominican, and 25% as Central or South American. The eleven variables used in this study include ethnicity, years spent in the US, gender, age, marital status, household income, employment, education, frequency of anxious feelings, frequency of depressive feelings, and level of non-specified psychological distress.

Measures

This study examines three outcomes: anxious feelings, depressive feelings, and psychological distress. Self-reported frequency of anxious feelings in the past year was scored on a five-point scale ranging from never (= 0) to daily (= 4), and self-reported frequency of depressive feelings in the past year contains the same five ordered response categories. Both variables were specified in two ways: the five-category scale ranging from never to daily, and a dichotomous measure based on reporting a frequency of weekly or higher. The third outcome, non-specified psychological distress in the past month, was measured using the Kessler-6 (K6) scale, which is a sum of six questions (e.g., “How often did you feel hopeless?”) scored on a five-point scale from none of the time (= 0) to all of the time (= 4). The K6 scale is widely used as an indicator of psychological distress, with studies showing that a score between 5 and 12 meets criteria for clinically relevant moderate psychological distress, and a score of 13 or higher indicates the presence of a diagnosable mental illness with considerable disability [22, 23]. This measure has been found to have high internal reliability among Latinx individuals with little evidence of nonequivalence between English and Spanish speaking adults [24, 25]. This variable was specified in two ways: a continuous measure ranging from 0 to 24, and a trichotomous measure ranging from no to low distress (= 0), moderate distress (= 1), and severe distress (= 2).

The independent variable of interest in this study measured migrants’ inclusion in five ethnic groups: Mexican, Puerto Rican, Cuban, Dominican, Central or South American. Demographic covariates include gender (coded as male = 0 and female = 1), age of respondents (continuous), and marital status (coded as not married = 0 and married = 1), all of which have been associated with mental wellbeing among migrants [13]. Socioeconomic status was measured by respondent’s annual household income (coded as $0 to $34,999 = 0, $35,000 to $49,999 = 1, $50,000 to $74,999 = 2, $75,000 to $99,999 = 3, and $100,000 and over = 4), employment (coded as unemployed = 0 and employed = 1), and education (coded categorically as less than high school = 0, high school graduate = 1, some college = 2, and bachelor’s degree and above = 3). Acculturation level was operationalized by using an ordinal variable measuring respondent’s years in the US (coded as < 1 year = 0, 1 < 5 years = 1, 5 < 10 years = 2, 10 < 15 years = 3, and ≥ 15 years = 4, and also coded dichotomously with ≤ 14 years = 0, and ≥ 15 years = 1 for a logistic regression, given prior findings of an association between increased time spent in the US and migrants’ mental wellbeing) [26, 27].

Preliminary analyses also tested for the effects of race using the measure available (five racial categories as defined by the US census). However, this variable was not found to be statistically significant and is excluded for subsequent analyses, partly for the sake of parsimony and largely because this dataset’s measurement of racialization is particularly deficient in relation to the US Latinx populationFootnote 3[7, 28]. Future research should focus on more accurately capturing the unique racialization experiences faced by Latinx migrants, particularly those of African and Indigenous descent.

Analysis

Crosstabulations were used to illustrate the weighted distributions of variables across Latinx ethnic groups with the US-born non-Latinx white population provided for reference (Table 1). Sample means of frequency of anxious feelings, depressive feelings, and level of non-specified psychological distress were calculated, and significance tests for differences among estimated means were conducted using the adjusted Wald statistic for the design-based F test. Bonferroni corrections were used for each pairwise comparison of means between the Latinx ethnic groups for each outcome variable. Weighted binary logistic regression and partial proportional odds (PPO) models with controls for demographic variables and socioeconomic variables were then used to compare the relative impact of a migrant’s membership within each ethnic group on odds of reporting frequent anxious feelings, frequent depressive feelings, and moderate and severe psychological distress. The predicted probabilities of reporting frequent feelings of anxiety, depression, and severe psychological distress were then calculated for each Latinx ethnic group. The data analysis was conducted in Stata 16 and the Stata survey (svy) command was applied to the regression models to account for the complex survey design of IPUMS Health Surveys.

Results

Table 1 presents weighted sociodemographic characteristics across Latinx ethnic groups. Mexicans and Central and South Americans reported the highest share of working-age adults (26–55), 72.5% and 69.2%, respectively, and the lowest shares of elderly adults. Mexicans reported the largest share of respondents, 55.7%, in the lowest education category. Cubans reported the largest share of respondents who received a bachelor’s degree or more at 23.2%. Central and South Americans reported the largest proportion of employed respondents at 69.8%, and Puerto Ricans reported the lowest at 46.8%. Puerto Ricans also reported the largest share of respondents with the lowest household income at 54.5%, followed by Mexicans at 49.8%.

Table 1 also presents the unadjusted prevalence rates for frequent feelings of anxiety, depression, and moderate and severe psychological distress. Puerto Ricans reported the highest unadjusted prevalence rate of 23.6% for frequent anxiety, followed by Cubans at 16.9%, the highest rate for frequent depression at 16.8%, followed by Dominicans at 9.1%, and the highest rate for severe psychological distress at 7.1%, also followed by Dominicans at 5.2%. To compare equality of means of each outcome variable across the different ethnic groups, significance tests were run for all pairwise differences among the means of the three outcome variables by ethnic groups using an adjusted Wald statistic for the design-based F test. The tests indicated significant differences in mean frequency of anxious feelings across ethnicity, F(4, 890) = 6.71, p < 0.001, mean frequency of depressive feelings, F(4, 890) = 11.13, p < 0.001, and mean level of non-specified psychological distress, F(4, 890) = 7.29, p < 0.001.

Table 2 presents the results of the binary logistic regression models that examine the associations between ethnic groups and the relative likelihood of reporting frequent feelings of anxiety and frequent feelings of depression, and the partial proportional odds (PPO) models that examine the relative likelihood of reporting moderate and severe psychological distress.

Table 1 Weighted sociodemographic characteristics of Latinx adults 18–85 born outside the US and US born non-Latinx whites.
Full size table

The PPO models can be interpreted similarly to the binary logistic models except the outcome variable has been collapsed into two categories: low distress is compared to moderate and severe distress in the first set of models, and then low and moderate distress is compared to severe distress in the next set of models. For each outcome variable: model 1 accounts only for country of origin; model 2 accounts for demographic characteristics including age, gender, marital status, and years of residence in the US; and model 3 accounts for socioeconomic status including education, employment, and household income. In all models, Puerto Ricans are the reference group.

In the first set of models across all outcome variables, membership in nearly every migrant group category was associated with significantly (p < 0.01) lower odds of reporting frequent feelings of anxiety, depression, and moderate and severe psychological distress (relative to Puerto Ricans). Mexicans and the aggregate group of Central and South Americans both reported the greatest differences in likelihood relative to Puerto Ricans. Mexicans were 43.5% less likely to report frequent anxiety and Central and South Americans were 41.3% less likely (model 3). Mexicans and Central and South Americans were 57.1% and 51.3% less likely, respectively, than Puerto Ricans to report frequent depression (model 6). Dominicans and Cubans reported the closest odds to Puerto Ricans across outcome variables (models 1-6 & 10-12), yet Dominicans were still statistically 40.6% less likely to report frequent anxiety, both were statistically less likely to report frequent depression (42% and 47.3%, respectively in model 6) and moderate/severe distress (41% and 41.3%, respectively in model 9). This pattern of differences in odds between migrants holds across outcome variables and model specifications.

All covariates behaved as expected in the literature. Women were 28–41% more likely to report frequent anxiety, frequent depression, and moderate/severe distress relative to men.

Table 2 Logistic regression models (in odds ratios) on reporting frequent anxiety, frequent depression, and severe non-specified psychological distress by ethnic groups, age, gender, marital status, acculturation levels, and SES.
Full size table

Married respondents were 30–40% less likely than unmarried respondents to report frequent anxiety, frequent depression, and moderate/severe distress. Migrants who have spent over 15 years in the US were 45-76% more likely to report frequent anxiety, frequent depression, and moderate/severe distress. Employed respondents were 40–57% less likely to report frequent anxiety, frequent depression, and moderate/severe distress. Figure 1 presents the predicted probabilities of reporting frequent feelings of anxiety, depression, and severe psychological distress, by ethnicity when all covariates are evaluated at their means. The predicted probabilities from the logistic regression analyses support the initial descriptive findings in Table 1 regarding the relationship between ethnicity and frequent feelings of anxiety, depression, and severe psychological distress.

Fig. 1
figure 1

Overlapping multicolor bar chart presenting the adjusted probabilities, along with the 95% confidence intervals, of reporting frequent feelings of anxiety, frequent feelings of depression, and moderate/severe psychological distress, by Latinx ethnicity when all covariates are evaluated at their means

Full size image

Discussion

In each pairwise comparison of ethnic groups, when controlling for demographic and socioeconomic characteristics, the same pattern emerges: Caribbean Latinx groups with their recent histories of US military intervention and occupation, and Puerto Ricans in particular, as the only respondents from a current colonial administration, report significantly higher predicted probabilities of frequent anxious feelings, depressive feelings, and moderate/severe psychological distress than migrants from countries with different colonial relations with the US. These results lend support to the notion that a gradient of exposure to US coloniality exists with respect to psychosocial well-being, with migrants from current colonial administrations (Puerto Ricans) on one end, migrants from recent colonial administrations (Cubans and Dominicans) holding middling positions, and migrants from states with less recent histories of US coloniality (Mexico) and neocolonial relationships (Central and South America) at the other end.

This gradient of exposure to the psychosocial harms of US coloniality, and the location of Puerto Ricans on it, can help explain recent findings indicating that the lifetime prevalence of psychiatric disorders among adult Puerto Ricans on the island are comparable to the overall rates of psychiatric disorders among adult Puerto Ricans in the US [29]. Puerto Ricans are exposed in the US as well as on the island, under the US colonial fiscal board and administration. Related, the higher rates of cyclical migration [30, 31] with flows of intermediate selection [32], reported among Puerto Ricans, continue to challenge explanations focused on immigrant selection mechanisms.

Limitations to this study include the operationalization of the variables measuring Latinx ethnicity and gender. The variable available to measure ethnicity had only one option available for any respondent who identified as Central or South American, severely limiting the findings for migrants from this region. Similarly, the variable measuring gender was operationalized within this dataset with only two values reflecting binary categories, failing to capture the ways that transgender and gender nonconforming individuals’ experiences differ from those of cisgender migrants (33–34). Additionally, given the cross-sectional nature of the data, causal mechanisms of the variables of interest were not able to be explored.

New Contribution to the Literature

This study highlights the importance of a disaggregated approach to the study of a population that experiences pan-ethnicization and minimization of intragroup variance in experiences and outcomes, and contributes to the literature in three ways. First, I extend the literature on Latinx mental wellbeing by using recently acquired data from 2010 to 2018 to update prior research. Second, I provide evidence that prior patterns concerning Puerto Rican migrants’ increased risk of diagnosable psychiatric disorders are reflected in contemporary data on self-reported feelings of mental wellbeing. Third, I propose and empirically test an innovative conceptual approach to the study of immigrant mental health rooted in varying degrees of exposure to coloniality. Mental health researchers and practitioners who prioritize developing strategies for reducing health disparities should consider the layered experiences that individuals with different ethnic backgrounds, socioeconomic positions and gendered and racialized bodies are faced with before, during, and after migration. Focusing on the unique colonial histories that position these migrants at intersecting systems of socio-political oppression is crucial for the promotion of health equity.